BackgroundIn developed countries, primary care physicians manage most patients with depression. Relatively few studies allow a comprehensive assessment of the decisions these doctors make in these cases and the factors associated with these decisions. We studied how general practitioners (GPs) manage the acute phase of a new episode of non-comorbid major depression (MD) and the factors associated with their decisions.Methodology/Principal FindingsIn this cross-sectional telephone survey, professional investigators interviewed an existing panel of randomly selected GPs (1249/1431, response rate: 87.3%). We used case-vignettes about new MD episodes in 8 versions differing by patient gender and socioeconomic status (blue/white collar) and disease intensity (mild/severe). GPs were randomized to receive one of these 8 versions. Overall, 82.6% chose pharmacotherapy; among them GPs chose either an antidepressant (79.8%) or an anxiolytic/hypnotic alone (18.5%). They rarely recommended referral for psychotherapy alone, regardless of severity, but 38.2% chose it in combination with pharmacotherapy. Antidepressant prescription was associated with severity of depression (OR = 1.74; 95%CI = 1.33–2.27), patient gender (female, OR = 0.75; 95%CI = 0.58–0.98), GP personal characteristics (e.g. history of antidepressant treatment: OR = 2.31; 95%CI = 1.41–3.81) and GP belief that antidepressants are overprescribed in France (OR = 0.63; 95%CI = 0.48–0.82). The combination of antidepressants and psychotherapy was associated with severity of depression (OR = 1.82; 95%CI = 1.31–2.52), patient's white-collar status (OR = 1.58; 95%CI = 1.14–2.18), and GPs' dissatisfaction with cooperation with mental health specialists (OR = 0.63; 95%CI = 0.45–0.89). These choices were not associated with either GPs' professional characteristics or psychiatrist density in the GP's practice areas.Conclusions/SignificanceGPs' choices for treating severe MD complied with clinical guidelines better than those for mild MD; GPs rarely recommended psychotherapy alone but rather as a complement to pharmacotherapy. Their decisions were mainly influenced by personal life experience and attitudes regarding treatment more than by professional characteristics. These results call into question the methods and content of continuing medical education in France about MD management.